Approximating rational points on surfaces (Q7008679)

The focus of the paper under review is the \textsl{approximation constant} \(\alpha(P,A)\) associated to a rational point \(P\) on a smooth projective variety \(X\) over a number field \(k\) with ample \(\mathbb{Q}\)-Cartier divisor \(A\). This notion generalizes the irrationality exponent associated to a real number.\N\NOne first defines \(\alpha(\{x_i\},P,A)\) for a sequence \(\{x_i\}\in X(k)\) of distinct points converging to \(P\) (with respect to a \(v\)-adic distance dist\((\cdot,\cdot)\) for a fixed place \(v\) of \(k\)) to be the infimum of the set \(\{\gamma\in\mathbb{R}\;|\;\mbox{dist}(P,x_i)^\gamma H_A(x_i)\mbox{ is bounded from above}\}\), where \(H_A\) is a Weil height associated to \(A\). The approximation constant \(\alpha(P,A)\) is then defined to be the infimum of \(\alpha(\{x_i\},P,A)\) for all such sequences (if no such sequence exists, then \(\alpha(P,A)\) is defined to be \(\infty\)).\N\NThe main point of interest in this paper is Conjecture 1.2, formulated by the second author in [\textit{D. McKinnon}, J. Algebr. Geom. 16, No. 2, 257--303 (2007; Zbl 1140.14016)], where a slightly different definition of approximation constant was given. The conjecture states that for \(\alpha(P,A)<\infty\) there exists a curve \(C\) on \(X\) such that \(\alpha(P,A)=\alpha(P,i^*A)\), where \(i\colon C\hookrightarrow X\) is the embedding. In other words, a sequence of rational points that `best approximates' \(P\) should lie on a curve \(C\).\N\NThe main result is Theorem 1.5, in which Conjecture 1.2 is proven for all split surfaces conditional on Vojta's Conjecture -- the latter allows one to reduce to proving Conjecture 1.2 for all \(P\) that are \textsl{essentially bounded} (Definition 1.3).\N\NPrevious results in this direction contain proofs of Conjecture 1.2 for many split rational surfaces, including all split smooth rational surfaces of Picard rank at most five, and some cases of non-split surfaces [\textit{D. McKinnon}, J. Algebr. Geom. 16, No. 2, 257--303 (2007; Zbl 1140.14016); \textit{D. McKinnon} and \textit{M. Roth}, Invent. Math. 200, No. 2, 513--583 (2015; Zbl 1337.14023); \textit{D. Castañeda}, PhD thesis, University of Waterloo (2019)]. In higher dimensions, Conjecture 1.2 was proven for some split toric varieties [\textit{Z. Huang}, Algebra Number Theory 15, No. 2, 461--512 (2021; Zbl 1472.14030)], and the second and third authors proved that Vojta's Conjecture implies Conjecture 1.2 for all \(\mathbb{Q}\)-factorial split toric varieties [\textit{D. McKinnon} and \textit{M. Satriano}, Trans. Am. Math. Soc. 374, No. 5, 3557--3577 (2021; Zbl 1464.14025)].\N\NThe proof of Theorem 1.5 uses tools from birational geometry and the Minimal Model Program. Using the same techniques, one could obtain similar results in higher dimensions, leading the authors to formulating Conjecture 1.7, a weaker version of Conjecture 1.2 for higher dimensions.